Our overall goal is to establish the basis for a new experimental paradigm for functional magnetic resonance imaging (fMRI) that makes possible quantitative measurement of the dynamics of the cerebral metabolic rate of oxygen metabolism (CMRO2) noninvasively in the human brain. Functional MRI methods based on blood oxygenation level dependent (BOLD) signal changes clearly have the potential to provide a window on CMRO2 dynamics, using simultaneous measurement of both the BOLD response to activation and the cerebral blood flow (CBF) response with a spiral dual-echo arterial spin labeling (ASL) technique. We and others have combined these tools in calibrated-BOLD studies to quantify changes in CMRO2, but these studies have focused on sustained changes in an approximate steady-state. The primary obstacle to extending these methods to measuring full CMRO2 dynamics is a physiological question: Do the dynamics of venous cerebral blood volume (CBVV) strongly differ from the dynamics of CBF? The key variable needed to estimate the dynamics of CMRO2 is the dynamics of the venous hemoglobin saturation, and the basic problem is that the BOLD effect depends primarily on changes in total deoxyhemoglobin, and thus also on the dynamics of venous blood volume. Dynamic measurements of CBF and BOLD signals provide sufficient information to estimate CMRO2 dynamics only if CBVV follows CBF. A primary example of this fundamental ambiguity of the BOLD signal is a long-standing issue in fMRI: is the post-stimulus undershoot of the BOLD signal a neural, vascular or metabolic effect? Despite considerable effort by many groups, there is still no clear answer, and the possibility of a dissociation of venous blood volume changes from CBF changes due to different dynamic time constants currently stands in the way of developing reliable tools for measuring CMRO2 dynamics. The motivation for this high risk/high gain proposal is that our recent studies of the effect of hyperoxia on the BOLD signal suggest a novel approach for addressing this primary physiological question, with a method that is specifically sensitive to CBVV. In addition, current models for the BOLD response and for analyzing the ASL experiment are essentially steady-state models, and these need to be expanded to include full dynamics. We will address these two basic limitations to measuring CMRO2 dynamics with two Aims. Aim 1: Extend our current modeling framework to include dynamics as well as potentially confounding physiologically variables, and use this to develop a Bayesian framework for estimating CMRO2 dynamics. Aim 2: Using the post- stimulus undershoot as a test case, use the hyperoxia approach to measure the dynamics of CBVV in human primary visual cortex in response to visual stimuli with varying duration and intensity. The endpoint will be a novel assessment of the dynamics of CBVV that will establish the feasibility of measuring the dynamics of CMRO2 for future applications in health and disease.